-
Strategic Management JournalStrat. Mgmt. J., 36: 1469–1485
(2015)
Published online EarlyView 22 August 2014 in Wiley Online
Library (wileyonlinelibrary.com) DOI: 10.1002/smj.2307Received 11
March 2013; Final revision received 16 June 2014
DOES PRODUCT MARKET COMPETITION FOSTERCORPORATE SOCIAL
RESPONSIBILITY? EVIDENCEFROM TRADE LIBERALIZATIONCAROLINE
FLAMMERIvey Business School, University of Western Ontario, London,
Ontario, Canada
This study examines whether product market competition affects
corporate social responsibility(CSR). To obtain exogenous variation
in product market competition, I exploit a quasi-naturalexperiment
provided by large import tariff reductions that occurred between
1992 and 2005 in theU.S. manufacturing sector. Using a
difference-in-differences methodology, I find that
domesticcompanies respond to tariff reductions by increasing their
engagement in CSR. This findingsupports the view of “CSR as a
competitive strategy” that allows companies to
differentiatethemselves from their foreign rivals. Overall, my
results highlight that trade liberalization is animportant factor
that shapes CSR practices. Copyright © 2014 John Wiley & Sons,
Ltd.
INTRODUCTION
Over the past decades, the rapid globalization ofthe world
economy has led to profound changesin the way companies operate. In
particular, tradeliberalization has contributed to an
unprecedentedincrease in the competitive pressure that U.S.
com-panies face from their foreign rivals (e.g., Bernard,Jensen,
and Schott, 2006a; Krugman, 1995; Krug-man, Obstfeld, and Melitz,
2012). This trendtowards lower trade barriers has spurred a large
lit-erature that studies how foreign competition
affectsproductivity (e.g., Bernard et al., 2006a), economicgrowth
(e.g., Frankel and Romer, 1999), as well associal and environmental
welfare (e.g., Edmondsand Pavcnik, 2005; Grossman and Krueger,
1993).While the latter focuses on social and environmentalwelfare
at the aggregate level, very little is known
Keywords: corporate social responsibility; product
marketcompetition; trade liberalization; competitive strategy;
dif-ference-in-differences*Correspondence to: Caroline Flammer,
Ivey Business School,University of Western Ontario, 1255 Western
Road, Office3351, London, Ontario N6G 0N1, Canada. E-mail:
[email protected]
Copyright © 2014 John Wiley & Sons, Ltd.
on how foreign competition affects firm-leveldecisions to invest
in corporate social responsi-bility (CSR), and in particular
whether domesticcompanies use CSR as a differentiation strategyto
compete against their foreign rivals. This papersheds light on this
question by theorizing andempirically testing how reductions in
importtariffs—which facilitate the entry of foreign com-petitors
into local markets—affect the social andenvironmental practices of
U.S. companies.
The concept of comparative advantage is a coretenet of
neoclassical trade theory (e.g., Heckscher,1919; Ohlin, 1933;
Ricardo, 1817) and strategicmanagement (e.g., Helfat and Peteraf,
2003; Hoo-ley, Broderick, and Moeller, 2006; Peteraf,
1993;Wernerfelt, 1984, 1995). In particular, incumbentcompanies can
sustain their competitive advantageby leveraging their resources
and capabilities inwhich they have a comparative advantage. In
thecontext of trade liberalization, domestic compa-nies have a
comparative advantage over foreigncompanies in their relationships
to local stake-holders. Hence, I argue that domestic companiesmay
respond to increased foreign competition bystrengthening their
relations with local consumers,
-
1470 C. Flammer
employees, and other stakeholders. Relatedly, theCSR literature
argues that companies can “do wellby doing good” as they may
benefit from higheremployee motivation, access to new market
seg-ments (such as “green” consumers), the more effi-cient use of
materials and energy, etc. (e.g., Hart,1995; Jones, 1995; Porter
and Kramer, 2006, 2011;Russo and Fouts, 1997). In line with these
argu-ments, I posit that increased foreign competitionmay foster
CSR since domestic companies are eagerto leverage their comparative
advantage to remaincompetitive.
Recent surveys are supportive of this theo-retical prediction.
Specifically, the surveys byAccenture and UNGC (2010) and MIT Sloan
Man-agement Review (2012) indicate that, in the face ofrising
global competition, over 90 percent of CEOssee sustainability as
critical for their company’scompetitiveness and future
success.1
Apart from these surveys, there is little evidenceon the impact
of foreign competition on CSR. Thisquestion is difficult to answer
empirically since tra-ditional measures of competition (e.g.,
import pene-tration) are likely endogenous with respect to CSR.In
other words, unobserved characteristics maydrive a spurious
correlation between the two. Forexample, long-term thinking CEOs
may be moreinclined to implement CSR initiatives. At the sametime,
they may self-select into non-competitiveindustries (e.g., because
the lower short-run pres-sure gives them more leeway in achieving
long-termobjectives). Another example is a reverse causal-ity
argument: companies could use CSR as a wayto influence competition.
In particular, incumbentcompanies may increase their CSR to preempt
entry
1 Relatedly, anecdotal evidence suggests that fiercer
competitionleads companies to increase their investment in CSR,
consistentwith the view of CSR as a competitive strategy. For
example,Seventh Generation’s CEO John Replogle argues that, in a
com-petitive environment where only the fittest survives, CSR is
key:“Sustainability is no longer optional. Companies that fail to
adoptsuch practice will perish. They will not only lose on a cost
basis,they will also suffer in recruiting employees as well as
attractingconsumers.” Furthermore, when referring to his former
company,Burt’s Bees, John Replogle argues: “Because we’ve trimmed
ouruse of electricity, water, waste, and most packaging inputs,
weare leaner and more competitive than most companies. … Burt’sBees
is a more competitive and profitable business BECAUSEwe embrace
sustainable practices” (Forbes, 2011, emphasis inoriginal). Along
similar lines, the declared objective of GeneralElectric’s
environmental CSR program “ecomagination” was toimprove GE’s
competitiveness. As GE’s CEO Jeffrey Immeltemphasizes: “We did it
from a business standpoint from Day 1,… it was never about
corporate social responsibility” (New YorkTimes, 2011).
of foreign firms. As these examples illustrate, find-ing a
correlation between, say, import penetrationand CSR would not
warrant a causal interpretation.2
To overcome this obstacle, I exploit aquasi-natural experiment
in the form of largeimport tariff reductions that occurred
between1992 and 2005 in the U.S. manufacturing sector.These tariff
reductions are substantial (tariff ratesdecreased by about 50% on
average), and henceprovide sharp exogenous shifts in the
competitivepressure that U.S. companies face from their
foreignrivals. To estimate the effect of these “treatments”on CSR,
I use a difference-in-differences approach.Specifically, if a firm
operates in an industry thatexperiences a tariff reduction (a
“treated” firm), Icompute the difference in CSR before and after
thetariff reduction. I then compare this difference withthe
corresponding difference at a “control” firm.Control firms are
matched to treated firms on thebasis of similar ex ante
characteristics.
Using this matched difference-in-differencesmethodology, I find
that tariff reductions lead tosignificant increases in CSR, as
measured by theKinder, Lydenberg, and Domini (KLD) index ofsocial
performance. This finding holds under alarge battery of robustness
checks including alter-native definitions of the treatment and
alternativematching procedures.
While tariff reductions provide plausibly exoge-nous variation
in competitive pressure from abroad,a potential concern is that
special interest groupsmay influence the outcome of trade policy.
Asa result, policymakers may reduce import tariffsbased on specific
industry characteristics (e.g., theymay lower tariffs in less
profitable industries asthey “give up” on them). If these
characteristicsare related to subsequent investments in CSR,my
results could be spurious. Nevertheless, thisconcern is mitigated
for two reasons. First, thematching algorithm ensures that treated
andmatched control firms are very similar ex ante,which alleviates
concerns that my results may bedriven by pretreatment differences
between treatedand control firms (e.g., in terms of
profitability).Second, I obtain similar results if I consider
only
2 A related strand of literature examines the associationbetween
domestic competition—as measured by the Herfindahl-Hirschman index
(HHI) of industry concentration—and CSR(Declerck and M’Zali, 2012;
Fernandez-Kranz and Santalo, 2010;Fisman, Heal, and Nair, 2006). As
with import penetration, HHI islikely endogenous with respect to
CSR. See the Discussion sectionfor more details.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Does Competition Foster CSR? 1471
the subset of tariff reductions that were part oflarge-scale
multilateral trade agreements estab-lished by the General Agreement
on Tariffs andTrade (GATT), World Trade Organization (WTO),and the
North American Free Trade Agreement(NAFTA). As Krugman et al.
(2012) argue, specialinterest groups are less likely to influence
tariffchanges resulting from multilateral trade agree-ments
compared to those that are negotiated on abilateral basis. Hence,
this subset of treatments isrelatively more exogenous with respect
to lobbyingpressure.
Finding that U.S. companies respond to highercompetitive
pressure from abroad by increas-ing their CSR is consistent with
the view thatCSR generates valuable resources that allowcompanies
to improve their competitiveness anddifferentiate themselves from
their foreign rivals.In auxiliary analyses, I further document
thatthis effect is stronger for companies operating inthe
business-to-consumer (B2C) sector—i.e., inindustries where the
purchasing decision is moresensitive to companies’ CSR engagement
(Lev,Petrovits, and Radhakrishnan, 2010). I also doc-ument that
companies focus their additional CSRinvestments on their core
stakeholders (customersand employees) as opposed to their other,
moreperipheral stakeholders (society at large andenvironment).
In the remainder of this paper, I develop thetheoretical
arguments in detail, describe the dataand methodology, present the
empirical results,and conclude by discussing the implications
andlimitations of my findings.
THEORY AND HYPOTHESIS
Relationship between foreign competitionand corporate social
responsibility
To derive theoretical predictions on the relationshipbetween
foreign competition and the CSR engage-ment of domestic companies,
I draw from differentstrands of literature.3 A long-standing
literature ineconomics examines the impact of trade liberaliza-tion
on economic growth (e.g., Frankel and Romer,
3 An activity is considered to be socially responsible if it
goesbeyond the firm’s maximization of its (single) bottom line
andlegal requirements and contributes to the social good (e.g.,
Davis,1973; McWilliams and Siegel, 2001).
1999), productivity (e.g., Bernard et al., 2006a), andemployment
(e.g., Wood, 1995). In particular, at thecore of neoclassical trade
theory is the concept ofcomparative advantage, according to which
coun-tries align their productive activities with their rel-ative
resource endowment (e.g., Heckscher, 1919;Ohlin, 1933; Ricardo,
1817). For example, asU.S. import tariffs decreased over the past
years,domestic companies faced increased global compe-tition from
low-wage countries such as India andChina. Given their relatively
higher wages, U.S.companies responded by shifting their
productionfrom labor-intensive products to more skill-
andcapital-intensive products (e.g., Bernard, Jensen,and Schott,
2006b; Pierce and Schott, 2012).
Relatedly, the strategic management literatureargues that
companies can sustain their competitiveadvantage by leveraging the
resources and capa-bilities in which they have a comparative
advan-tage (e.g., Helfat and Peteraf, 2003; Hooley et al.,2006;
Peteraf, 1993; Wernerfelt, 1984, 1995). In thecontext of trade
liberalization, domestic companieshave a comparative advantage over
foreign compa-nies in their relationship with local
stakeholders,while it may be difficult for them to compete ona cost
basis.4 Accordingly, I argue that domesticcompanies may respond to
increased foreign com-petition by strengthening their relations
with localconsumers, employees, and other stakeholders. In asense,
by stepping up their social and environmentalinitiatives, companies
can differentiate themselvesand establish a “soft” trade barrier
disadvantagingtheir foreign competitors.
The potential value of strengthening firms’ rela-tions with
their stakeholders is also emphasized inthe CSR literature. For
instance, Freeman’s (1984)stakeholder theory suggests that
companies shouldconsider the interests of a broader group of
stake-holders. Several extensions of stakeholder theoryhave been
proposed (for a review, see Agle et al.,2008). In particular,
instrumental stakeholder the-ory (e.g., Jones, 1995) holds that CSR
efforts canbe instrumental in obtaining necessary resourcesor
stakeholder support. Similarly, companies mayengage in CSR in order
to improve their efficiencyand enhance, e.g., their reputation,
brand, and trust(e.g., Barney, 1991; Hart, 1995; Porter, 1991;
Russoand Fouts, 1997). This argument is related to Porter
4 The cost advantage of foreign rivals is likely one of the
rationalesunderlying the use of import tariffs in the first place
(e.g., Gros,1987; Helpman and Krugman, 1989).
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
1472 C. Flammer
and Kramer (2006, 2011) who emphasize the strate-gic importance
of considering a broader businessenvironment and creating “shared
value” for bothsociety and the company. The creation of
sharedvalue—as opposed to only social (i.e., philan-thropic)
value—is integral to a company’s max-imization of long-term
shareholder value and itscompetitiveness in the global market
place.
In sum, the above arguments imply that CSRallows domestic
companies to improve theircompetitiveness and differentiate
themselves fromtheir foreign rivals. Accordingly, companies
facingfiercer competition from abroad may respond byincreasing
their investment in CSR. Hence, I posita positive causal
relationship between foreigncompetition and CSR:
Hypothesis 1: An exogenous increase in foreigncompetition leads
to an increase in CSR.
Naturally, the alternative hypothesis is that anincrease in
foreign competition leads to a decrease(or no change) in CSR, which
would be in line withthe literature that sees companies’ social
engage-ment as an inefficient use of resources. For
example,Friedman’s shareholder theory (Friedman, 1962,1970) views
social responsibility as an unnecessarycost of doing business.
Accordingly, addressingsocial issues reduces the company’s profits
and isakin to a transfer from shareholders to stakeholders.A
similar argument is made, e.g., in Elhauge (2005)who argues that
CSR policies involve “sacrificingcorporate profits in the public
interest” (p. 733). Inthe spirit of this literature, an increase in
competi-tive pressure may stifle CSR, since it reduces
firms’profits and hence the amount of resources that canbe
transferred to stakeholders.
DATA
Reduction of import tariff rates
To measure increases in foreign product marketcompetition, I use
industry-level import tariff datacompiled by Feenstra (1996),
Feenstra, Romalis,and Schott (2002), and Schott (2010). These
dataare available at the four-digit SIC (Standard Indus-try
Classification) level for the U.S. manufacturingsector (SIC
2000–3999) from 1972 to 2005. Foreach four-digit SIC industry and
year, I compute thead valorem tariff rate, which is the ratio of
duties
collected by U.S. Customs to the free-on-boardvalue of
imports.
Tariff rates fluctuate from year to year. However,the typical
tariff change is very small and econom-ically unimportant. To
circumvent this limitation, Ifollow common practice in the
economics literatureand consider only “large” tariff reductions,
i.e., tar-iff reductions that are above a certain threshold
(e.g.,Fresard, 2010; Fresard and Valta, 2014; Lileeva andTrefler,
2010; Trefler, 2004). Specifically, I followFresard (2010) and
Fresard and Valta (2014) andqualify a tariff rate reduction in a
given industryyear as large if it is at least three times larger
thanthe average annual (absolute) change in tariff rate inthe same
industry across all years. The choice of thethreshold is immaterial
for my analysis. In robust-ness checks, I show that my results also
hold if Iconsider alternative cutoffs such as tariff reductionsthat
are two or four times the average.
There are 91 such large tariff reductions from1972 until 2005;
the first one occurring in 1975,the last one in 1998. Since the
objective of thispaper is to study how import tariff reductions
affectCSR, and given that CSR data from the KLDdatabase are
available from 1991 onward, I onlyconsider tariff reductions that
occurred as of 1992.(Dropping events occurring in 1991 is due to
thedifference-in-differences specification that requiresat least
one year of CSR data in the year preced-ing the tariff reduction.)
This criterion leaves mewith a final set of 34 large tariff
reductions, whichare provided in Table S1. For each event, Table
S1reports the year of the tariff reduction, the four-digitSIC code,
a short description of the industry, andwhether the tariff
reduction was implemented aspart of multilateral trade agreements
established bythe GATT, WTO, or NAFTA. The latter informa-tion is
obtained from the U.S. International TradeCommission.5
These events correspond to an average decreasein tariff rates by
about 50 percent (on average, thetariff rate drops from 2.6% in the
year precedingthe event to 1.3% in the year following the
event).Accordingly, the treatments considered in this studyprovide
sharp increases in competitive pressurefaced by U.S. companies. For
more details about
5 The sample period considered in this study has witnessed
adecreasing trend in import tariffs. Accordingly, there are onlytwo
instances of large tariff rate increases. This prevents me
fromconducting the reverse analysis, i.e., studying whether
companiesadjust their social engagement following a decrease in
productmarket competition.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Does Competition Foster CSR? 1473
the economic significance of the treatments, see theMethodology
section.
Firm-level data
The accounting data are obtained from Standard &Poor’s
(S&P) Compustat; the CSR data are fromthe KLD database. KLD is
an independent socialchoice investment advisory firm that compiles
rat-ings of how companies address the needs of theirstakeholders.
During the relevant sample period, theKLD database consists of all
companies listed inthe S&P 500 Index as well as companies
listed in theDomini 400 Social Index, which includes mainlylarge
and mid-sized companies (see Domini SocialInvestments, 2013). KLD
ratings are widely usedin CSR studies (e.g., Berman et al., 1999;
Deckop,Merriman, and Gupta, 2006; Graves and Waddock,1994).
The KLD database contains social ratings ofcompanies along
several dimensions includingcommunity, diversity, employee
relations, envi-ronment, human rights, product quality,
corporategovernance, and whether firms’ operations arerelated to
alcohol, firearms, gambling, tobacco,nuclear power, and military
contracting. To con-struct a composite KLD index, I sum up
allstrengths along these dimensions.6 In auxiliaryanalysis, I also
consider four sub-indices of thiscomposite index (see the Results
section).
METHODOLOGY
Difference-in-differences
To study whether an increase in competitivepressure from abroad
affects CSR, I use adifference-in-differences methodology based
on
6 A few of the specific strengths are not surveyed every yearin
the KLD database, which could lead to inconsistencies in
themeasurement of CSR over time. However, I have verified that
Iobtain similar results if, instead of using the full index, I
onlyinclude those strengths that are surveyed in all years from
1991to 2005. In addition to CSR strengths, the KLD data also
containa list of CSR concerns. Accordingly, an alternative approach
is toconstruct a “net” KLD index by subtracting the concerns from
thestrengths. However, recent research suggests that this
approachis methodologically questionable. Because KLD strengths
andconcerns lack convergent validity, using them in conjunction
failsto provide a valid measure of CSR (e.g., Johnson-Cramer,
2004;Mattingly and Berman, 2006). For this reason, my analysis
relieson the composite index of KLD strengths (for a similar
approach,see, e.g., Kacperczyk, 2009).
the 34 large tariff reductions listed in Table S1(treatments).
Specifically, I compare the differencein KLD index before and after
the treatment forfirms in industries that experience large
tariffreductions (treatment group) with the correspond-ing
difference for firms that are not affected bythe tariff reductions
but are otherwise similar(control group). In the following, I
describe howthe treatment and control groups are constructed.
Treatment group
The treatment group consists of all firms that oper-ate in a
four-digit SIC industry that experiences alarge tariff reduction
and have coverage in Compu-stat and the KLD database at least one
year beforeand one year after the tariff reduction. The 34
largetariff reductions yield a sample of 254 treated firmsthat
satisfy these criteria.
Control group
To construct a sample of firms that are similar tothe treated
firms (except for the tariff reduction), Imatch each treated firm
to a control firm on the basisof industry- and firm-level
characteristics using thefollowing procedure.
First, since the treatments are at the industrylevel, matching
control firms based on the samefour-digit SIC industry is not
possible. Instead, anatural approach is to match control firms
basedon a broader industry sector such as one-, two-,or three-digit
SIC codes (excluding four-digit SICindustries that are treated). In
my baseline analysis,I require that the control firm operates in
the sametwo-digit SIC industry and produces the same typeof goods
(consumer versus intermediate goods).7
This approach balances two concerns. On one hand,the industry
partition needs to be sufficiently finegrained so that industry
characteristics are similar.On the other hand, the industry
partition needsto be broad enough so that the pool of
potentialcontrol firms for the matching based on
firm-levelcharacteristics is sufficiently large.
Second, out of the remaining candidates, I selectthe nearest
neighbor on the basis of six firm-level
7 The partition of four-digit SIC industries into consumer
versusintermediate goods is obtained from Lev et al. (2010: 188).
Iobtain very similar results if the industry matching is done
solelybased on two-digit SIC codes. In robustness checks, I
discussalternative matching procedures.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
1474 C. Flammer
Table 1. Summary statistics for treated and matched control
firms
Observations Mean25th
percentile50th
percentile75th
percentilep -value(t-test)
p-value(KS-test)
Panel A. Matching characteristicsKLD index Treated 254 1.751
0.000 1.000 3.000 0.918 0.816
Control 254 1.759 0.000 1.000 3.000Log(assets) Treated 254 7.954
7.064 7.791 9.113 0.275 0.173
Control 254 8.138 6.847 8.025 9.574Market-to-book Treated 254
2.092 1.419 1.655 2.310 0.267 0.353
Control 254 2.262 1.329 1.594 2.599ROA Treated 254 0.069 0.038
0.063 0.096 0.464 0.596
Control 254 0.066 0.026 0.061 0.103Cash/assets Treated 254 0.089
0.027 0.053 0.113 0.254 0.620
Control 254 0.098 0.028 0.047 0.138Leverage Treated 254 0.132
0.065 0.132 0.198 0.990 0.795
Control 254 0.132 0.047 0.135 0.198Panel B. Industry
characteristicsImport tariff rate Treated 254 0.028 0.022 0.029
0.033 0.483 0.395
Control 254 0.027 0.017 0.025 0.034Import penetration Treated
254 0.226 0.060 0.181 0.414 0.940 0.861
Control 254 0.228 0.060 0.202 0.402HHI Treated 254 0.316 0.125
0.231 0.516 0.747 0.594
Control 254 0.310 0.155 0.236 0.503
characteristics: KLD index, size, market-to-bookratio, return on
assets (ROA), cash holdings, andleverage ratio, all computed as
average in thethree years preceding the tariff reduction (using
pre-treatment values ensures that the matching charac-teristics are
not affected by the treatment itself).8
The nearest neighbor is the firm with the low-est Mahalanobis
distance to the treated firm acrossthese six matching
characteristics.9
This matching procedure ensures that controlfirms are as similar
as possible to the treated firmsex ante. In particular, using the
KLD index asa matching characteristic ensures that treated
andcontrol firms have similar CSR strengths prior to thetreatment.
Using measures of profitability (ROA)and growth opportunities
(market-to-book) rules
8 The last five characteristics are obtained from Compustat.
Size isthe natural logarithm of the book value of assets;
market-to-bookratio is the ratio of the market value of equity to
the book value ofequity; ROA is the ratio of income before
extraordinary items tothe book value of assets; cash holdings is
the ratio of cash andshort-term investments to the book value of
assets; leverage isthe ratio of long-term debt to the book value of
assets. These fivecharacteristics are commonly used in the
economics and financeliterature to construct a set of comparable
firms (see, e.g., Almeidaet al., 2012; Fresard and Valta, 2014).9
Formally, the Mahalanobis distance 𝛿 between treated firm iand
candidate firm j is given by 𝛿 = [(Xi −Xj)’ 𝚺−1 (Xi −Xj)]
1∕2 ,where X is a (6× 1) vector containing the six matching
variablesand 𝚺 is the (6× 6) covariance matrix of these six
variables.
out concerns that the treated firms may be lessprofitable or
operate in declining industries. Usingsize, cash holdings, and debt
capacity (leverage) fur-ther addresses the possibility that
differences alongthese characteristics may affect future CSR
invest-ments (e.g., through the ease of raising capital). Insum,
the control firms provide a counterfactual forwhat would happen at
the treated firms absent anyincrease in foreign competition. Since
each treatedfirm is matched to one control firm, the final sam-ple
consists of 508 companies (254 treated firms and254 matched control
firms).
To illustrate the similarity between treated andcontrol firms,
Table 1 reports descriptive statisticsfor the six matching
characteristics, as well threeindustry characteristics that capture
the degree ofcompetition in the four-digit SIC industry of
thetreated and control firms. These three character-istics are the
import tariff rate, import penetra-tion, and the
Herfindahl-Hirschman index (HHI) ofindustry concentration, all
computed as average inthe three years preceding the tariff
reduction.10 For
10 Import penetration is computed as the total imports divided
bythe sum of total domestic production plus imports minus exportsat
the four-digit SIC level. The data on import penetration
areobtained from Peter Schott’s website and are described in
Feenstra(1996) and Feenstra et al. (2002). The
Herfindahl-Hirschmanindex of (domestic) industry concentration is
computed as the sum
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Does Competition Foster CSR? 1475
each characteristic, the table reports means, medi-ans, 25th,
and 75th percentiles for both the 254treated firms and the 254
matched control firms. Inthe last two columns, the table further
reports thep-value of the difference-in-means test (t-test)
andKolmogorov-Smirnov test (KS-test), respectively.11
As is shown, treated and control firms are verysimilar along all
these characteristics. In particular,the null of equal means cannot
be rejected (withp-values ranging from 0.25 to 0.99). Neither
canthe null of equal distributions (p-values from 0.17to 0.86).
Overall, the statistics in Table 1 confirmthat control firms are
very similar to treated firms,and hence likely provide a reliable
counterfactualof how treated firms would behave absent the
tariffreductions.
For each treated firm and each matched controlfirm, I compute
the difference in the firm’s aver-age KLD index in the three years
following the tar-iff reduction minus the firm’s average KLD
indexin the three years preceding the tariff reduction.12 Idenote
this difference by ΔKLDit, where i indexesthe company and t indexes
the year of the tariffreduction. While I focus on three years
before andafter the tariff reductions in the baseline
specifica-tion, my results are not sensitive to the choice ofthe
treatment window. Specifically, I have verifiedthat my results are
robust if I use one, two, four, orfive years before and after the
treatment.
Having computed ΔKLDit for the treated andmatched control firms,
I can measure the effectof tariff rate reductions on CSR by
estimating thefollowing regression:
Δ KLDit = 𝛼t + 𝛽 × Tariff Reductionit+ 𝛾 ′Xit + 𝜖it,
of squared market shares of all companies in a given four-digit
SICindustry. Market shares are computed from Compustat based
onfirms’ sales.11 The KS-test is a nonparametric test of the null
hypothesis ofidentical distributions. The underlying test statistic
quantifies thedifference between the empirical distribution of the
variable ofinterest in the treatment group and its empirical
distribution inthe control group (for details, see Hollander and
Wolfe, 1999:178–186).12 The sample of treated and control firms is
constructed byrequiring that each firm has KLD coverage at least in
the yearbefore and the year after the treatment. In cases where
KLDdata are not available for the full three years before or after
thetreatment, the respective average is computed on the basis ofthe
nonmissing years. My results are virtually identical if I
onlyinclude firms with the full three years of KLD data before and
afterthe treatment.
where 𝛼t are year fixed effects, Tariff Reduction is adummy
variable (treatment dummy) that equals onefor treated firms and
zero for matched control firms,X is the vector of control
variables, which includesthe six characteristics used to construct
the matchedcontrol group (KLD index, size, market-to-bookratio,
ROA, cash holdings, and leverage ratio,all computed as average in
the three years pre-ceding the tariff reduction), and 𝜖 is the
errorterm. I cluster standard errors at the four-digitSIC industry
level. (I obtain similar results ifinstead I cluster standard
errors at the year level,at both the year and industry level, at
the two-digitSIC level, or if I use
heteroskedasticity-robuststandard errors.) The coefficient of
interest is 𝛽,which measures the difference in ΔKLD betweentreated
firms and matched control firms (i.e.,
thedifference-in-differences). In other words, it mea-sures the
effect of tariff reductions on the KLDindex accounting for
contemporaneous changes inthe KLD index at otherwise similar firms
that do notexperience such tariff reductions.
Validity of the identification strategy
To be valid, my identification strategy needs to ful-fill two
requirements. First, the treatments—i.e., thelarge import tariff
reductions—need to trigger rele-vant changes in the competitive
pressure that U.S.companies face from their foreign rivals.
Secondthe treatments need to be exogenous with respect toCSR. In
the following, I discuss both requirements.
Relevance of large import tariff reductions
Import tariffs have decreased gradually over thepast decades
(Bernard et al., 2006a; Feenstra, 1998;Krugman, 1995; Krugman et
al., 2012). This trendis visible in the figure provided in Figure
S1, where Iplot the evolution of import tariff rates in treated
andcontrol industries (i.e., the four-digit SIC industriesof the
treated and control firms, respectively). Ascan be seen, five years
prior to the treatment, importtariff rates were about 3.2 percent
in both controland treated industries. In control industries,
importtariffs decrease by about 0.2 percentage points everyyear.13
Import tariffs decrease at a similar pace
13 This decrease is representative of the average change in
importtariff rates across all manufacturing industries during the
sampleperiod. The corresponding average is −0.2 percentage point
peryear as well.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
1476 C. Flammer
in treated industries except in the year of thetreatment when
the tariff rate drops by half, from2.6 to 1.3 percent (i.e., a
reduction by 1.3 percentagepoints). This large reduction generates
a significantwedge between treated and control industries.
Thiswedge is persistent in the five years following
thetreatment.
In principle, a given industry can be treatedseveral times.
Nevertheless, such multiple treat-ments do not occur during the
sample period (seeTable S1). Moreover, none of the treatments
isreversed. This is in line with previous literaturedocumenting
that large increases in import tar-iffs are fairly rare (e.g.,
Fresard and Valta, 2014;Krugman et al., 2012). From an
identification per-spective, the absence of reversals and
multipletreatments is appealing, as it mitigates concerns thatmy
results may be contaminated by post-treatmentinterventions.
To interpret the magnitude of the large importtariff reductions,
it is helpful to benchmark themwith the Canada-U.S. Free Trade
Agreement (FTA)of 1989. Trefler (2004) reports that the passageof
the FTA lowered the average tariff rate forCanadian products from
four percent in 1988 toabout three percent in 1990, i.e., a
decrease byone percentage point. The FTA is commonly viewedas a
sizable event that substantially increased thecompetitive pressure
faced by U.S. companies (e.g.,Clausing, 2001; Trefler, 2004). In
terms of themagnitude, the average treatment in my
sample—areduction by 1.3 percentage points in import tariffrates—is
close to the tariff reduction brought aboutby the FTA.
A related point is whether managers pay closeattention to import
tariffs. Anecdotal evidence sug-gests that managers are indeed
sensitive to importtariff reductions. For example, when referring
to therecently proposed Trans-Pacific Partnership tradedeal, the
CEO of New Balance Athletic Shoe Inc.noted: “A rapid reduction of
the existing [tariff]agreements would put our factories here at
signif-icant risk” (Wall Street Journal, 2013). To obtainmore
systematic evidence on managers’ attention toimport tariffs, I
follow the approach of Fresard andValta (2014) and conduct a
textual analysis of theManagement’s Discussion and Analysis
(MD&A)section of the companies’ 10-K filings (i.e., theirannual
reports). Since 10-K filings are electron-ically available on the
SEC website from 1997onward, I conduct this analysis for the subset
ofcompanies that are treated as of 1997. Specifically,
I search the MD&A section for keywords pertainingto
“increasing competition.”14 I find that followinglarge import
tariff reductions, treated companies are38 percent more likely to
talk about increased com-petitive pressure, while the corresponding
increaseis merely seven percent for matched control firms.
Exogeneity of large import tariff reductions
My identification strategy relies on the assumptionthat large
import tariff reductions are exogenouswith respect to CSR. In the
following, I discusspotential identification concerns and describe
howmy matched difference-in-differences specificationis helpful in
addressing them.
Political economy of tariff changes. Tariffchanges are often the
result of a long negotiationprocess that may involve various
interest groups(e.g., Frye and Mansfield, 2004; Grossman
andHelpman, 1995; Henisz and Mansfield, 2006; His-cox, 2002; Mayer,
1981; Rogowski, 1989). Hence,a potential concern could be that
policymakersreduce import tariffs based on specific
industrycharacteristics that are related to subsequentinvestments
in CSR. For example, it could be thatpoliticians lower tariffs in
declining industriesas they “give up” on them. Or it could be
thatpolicymakers reduce tariffs in industries thathave become
sufficiently strong to face increasedcompetition from abroad. Or it
could be that importtariff reductions are more likely to occur in
thoseindustries where import tariff rates have beenunusually high.
In all these scenarios, there aresystematic differences between
treated and controlfirms (e.g., in terms of profitability) prior to
thetreatment. If these differences affect subsequentinvestments in
CSR, my results could be spurious.
The matching algorithm ensures that controlfirms are very
similar to treated firms prior to thetreatment, which alleviates
concerns that preexist-ing differences may affect my results. For
example,if large import tariff reductions are more likelyto occur
in declining industries, a potential con-cern is that treated firms
might be less profitablethan control firms. Nevertheless, as can be
seenin Table 1, there is no significant difference in
14 More precisely, I search for the word “increasing” or one if
itssynonyms (such as “increased,” “higher,” “greater,”
“intensified,”or “intensification”) appearing besides the word
“competition” (orvariations thereof).
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Does Competition Foster CSR? 1477
profitability (ROA) prior to the treatment. Sim-ilarly, the
evidence provided in Table 1 showsno preexisting difference in
growth opportunities(market-to-book ratio), financing policies
(cashholdings, leverage ratio), CSR (KLD index), aswell as the
degree of competition (Herfindahl-Hirschman index, import
penetration, import tariffrate).15
An additional way to address the above concern isby focusing on
large import tariff reductions that areharder to influence by
special interest groups. Whilespecial interest groups may influence
the outcomeof bilateral trade agreements, doing so is much
moredifficult for multilateral trade agreements such asthose that
were established by the GATT, WTO, orNAFTA (e.g., Fresard and
Valta, 2014; Krugmanet al., 2012; Ornelas, 2005). Indeed, the
participa-tion of multiple countries makes negotiations
moredifficult and hence limits the ability of governmentofficials
to give in to lobbying pressure. Moreover,international trade
institutions impose rules and for-mal obligations that restrict the
influence of specialinterest groups. Accordingly, import tariff
reduc-tions that were introduced as part of the GATT,WTO, and NAFTA
can be viewed as relatively moreexogenous compared to those
resulting from bilat-eral agreements. In robustness checks, I show
thatmy results are similar if I only consider this subsetof
treatments.
Anticipation of import tariff changes. A relatedconcern is that
companies may anticipate thetreatment and adjust their CSR
accordingly. Forexample, it could be that, in anticipation of
futurecompetitive pressure, companies momentarilycut nonmarket
activities (e.g., CSR) and focuson market activities. As companies
resume theirnonmarket activities following the treatment,my
estimates would capture a spurious increasein CSR after the
treatment. Nevertheless, thisconcern is unlikely to explain my
results, for tworeasons. First, the matching
algorithm—whichincludes the (pretreatment) KLD index as one ofthe
matching characteristics—ensures that there is
15 This evidence does not imply that import tariff changes
areunrelated to, e.g., profitability, investment opportunities, or
thedegree of competition. What it shows is that control firms
arevery similar to treated firms along these characteristics,
whichmitigates concerns that preexisting differences—such as
thosereflecting the political economy of import tariff
changes—mayaffect my results. See, e.g., Krugman et al. (2012) for
a discussionof the determinants of import tariff changes.
no preexisting difference in the KLD index in thethree years
preceding the treatment (see Table 1).Second, Figure 1 shows that
(1) the evolution ofthe KLD index is virtually identical among
treatedand control firms in the five years preceding thetreatment,
and (2) treated companies do not reducetheir KLD index in the
pretreatment years.
Related industries. Another potential concernis that a tariff
reduction in one industry mayaffect companies in related industries
(e.g., sup-pliers), even if the latter do not experience areduction
in tariff rates. If companies from suchindustries happen to be in
the control sample,the requirement that control firms be
unaffectedby the treatment would be violated. While it isunclear
how such industry spillovers would biasmy results, I show in
robustness checks that myresults are unchanged if I require control
firms tooperate in industries that are unrelated to those ofthe
treated firms. To measure relatedness acrossindustries, I use the
1992 input-output matrix ofthe Bureau of Economic Analysis and
computeinterindustry relatedness following the procedurein Fan and
Lang (2000). Industries are said to berelated if their relatedness
coefficient is larger thanfive percent.
Advertising. Finally, the KLD index may corre-late with
advertising and public relations expenses.In particular, it could
be that companies advertisetheir existing CSR more aggressively
following anincrease in foreign competition. If KLD analysts
areinfluenced by advertising campaigns in assessinga company’s
social performance, my results couldmerely reflect a change in
advertising behavior asopposed to an actual increase in CSR. To
mitigatethis concern, I show in robustness checks that myresults
are very similar if I control for contempora-neous changes in
advertising expenses (defined asthe ratio of advertising expenses
to total assets fromCompustat).
RESULTS
Main results
The main results are presented in Table 2. In allregressions,
the dependent variable is the change inKLD index three years after
compared to three yearsbefore the treatment. In Model 1, the
regres-sion only includes the tariff reduction dummy as
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
1478 C. Flammer
– 0.60
– 0.40
– 0.20
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
2.20
2.40
2.60
2.80
– 5 – 4 – 3 – 2 – 1 1 2 3 4 5
KL
D-in
dex
Year relative to treatment
Treatment group
Control group
Difference between treatment and control group (95% confidence
interval within dotted lines)
Figure 1. Evolution of KLD index in control and treatment
group
Table 2. Do import tariff reductions lead to higher CSR?
Δ KLD Δ KLD Δ KLD Δ KLDDependent variable Model 1 Model 2 Model
3 Model 4
Tariff reduction 0.402*** 0.403*** 0.363*** 0.316***(0.090)
(0.089) (0.085) (0.091)
Control variables No No Yes YesYear fixed effects No Yes Yes
YesRegression type OLS OLS OLS MedianR-squared 0.04 0.05 0.12
0.05Observations 508 508 508 508
Standard errors are in parentheses. All tests are
two-tailed.*p< 0.10; **p< 0.05; ***p< 0.01.
explanatory variable. In Model 2, I also include yearfixed
effects. In Model 3, I further include firm-levelcontrols (KLD
index, size, market-to-book ratio,ROA, cash holdings, and leverage,
all measured asaverage in the three years preceding the tariff
reduc-tion). Finally, in Model 4, I use a median (mean
absolute deviation) regression instead of ordinaryleast squares
(OLS).16 For each specification, the
16 Since clustering techniques are not available for
medianregressions, standard errors in Model 4 are
block-bootstrapped atthe four-digit SIC level using 500 bootstrap
samples.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Does Competition Foster CSR? 1479
table reports the coefficient on the tariff reductiondummy and
its standard error in parentheses. Ascan be seen, the coefficient
on the tariff reduc-tion dummy is very stable regardless of the
spec-ification.17 More precisely, it lies between 0.316and 0.403
and is always highly significant. Thisimplies that, in the three
years following the tariffreduction, companies increase their
social perfor-mance by about 0.3–0.4 KLD strengths—looselyspeaking,
companies are implementing 0.3–0.4CSR initiatives. While this
effect may seem modestin absolute terms, it is quite substantial in
relativeterms. Since the average number of KLD strengthsprior to
the treatment is 1.75 (see Table 1), thisimplies that the CSR
engagement of companiesincreases by about 18–23 percent.
To provide more perspective on the effect of tariffreductions on
CSR, Figure 1 plots the evolution ofthe KLD index in the treatment
(black solid line)and control group (black dashed line) five
yearsbefore and after the treatment, as well as the differ-ence
between the two (gray solid line) with the cor-responding 95
percent confidence interval (dottedlines).18 This figure provides
four insights. First, theKLD index is trending upward in both the
controland treatment groups. This is consistent with previ-ous
evidence showing that companies are increasingtheir CSR activities
over time (see, e.g., Flammer,2013), and underscores the importance
of using acontrol group—not accounting for changes in CSRat the
control group would overstate the effect of tar-iff reductions on
the KLD index, as it would capturesome of the time trend. Second,
there is no appar-ent difference in the KLD index in the five
yearspreceding the treatment. Third, following the treat-ment, the
two curves diverge: treated firms increasetheir KLD index
substantially more compared tomatched control firms. Fourth, Figure
1 sheds lighton the dynamics of the treatment effect. Compa-nies
start increasing their CSR in the first yearfollowing the tariff
reduction. However, it is onlyafter two years that the effect
becomes substantialand significant at the five percent
level—arguably,
17 Throughout the analysis, the inclusion of controls is
immaterialfor my results. This is to be expected given that the
variables usedas controls are the same as the matching
characteristics reportedin Table 1.18 Each point in the figure
represents the average KLD indexamong all firms in the respective
group (or the difference betweenthe two). In case a company does
not have KLD coverage ina given year, the average is based on the
remaining firms withnonmissing KLD data.
it may take some time for companies to decideupon and implement
the appropriate CSR program.Subsequently, the difference remains
significant andsomewhat stable in magnitude.
Robustness checks
I perform several robustness checks that addresspotential
concerns. All these robustness checks areprovided in Table S2.
First, I show that my results are not sensitiveto the coding of
the large import tariff reductions.In the baseline analysis, a
tariff reduction is codedas large if it is at least a threefold of
the average(absolute) tariff change in the industry. I
obtainsimilar results if a two- or fourfold cutoff is usedinstead
(Models 1 and 2 in Table S2).19
Next, I show that my results are robust to alter-native
definitions of the matched control group.Specifically, I obtain
similar results if all match-ing characteristics are measured three
years priorto the treatment (as opposed to the average ofthe three
years preceding the treatment), if con-trol firms are required to
operate in the samethree-digit SIC industry as treated firms, if
controlfirms are required to be located in the same stateas treated
firms (using the state of headquarters’location from Compustat), or
if I require that con-trol firms operate in industries that are not
verti-cally related to the treated industries (Models 3–6 inTable
S2).
Finally, I show that my results are similar if Ionly consider
large import tariff reductions thatwere established by multilateral
trade agreements(GATT, WTO, or NAFTA), or if I control
forcontemporaneous changes in advertising expenses(Models 7 and 8
in Table S2).
Auxiliary analysis
In Table 3, I provide auxiliary evidence thatis indicative of
potential mechanisms throughwhich CSR may improve companies’
ability tocompete with their foreign rivals. Note that thisevidence
is merely suggestive as it is open toalternative interpretations
(see the Discussionsection).
19 Interestingly, the coefficient is smaller for the twofold
cutoff(0.251) and larger for the fourfold cutoff (0.504), compared
to thecoefficient of 0.363 for the threefold cutoff. This pattern
suggeststhat the increase in CSR is monotonic in the extent to
whichforeign product market competition increases.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
1480 C. Flammer
Table 3. Auxiliary analysis
B2C sector KLD sub-indices for stakeholder groups
Δ KLD Δ KLD Δ KLD Δ KLD Δ KLDEmployees Consumers Environment
Society at large
Dependent variable Model 1 Model 2 Model 3 Model 4 Model 5
Tariff reduction 0.250*** 0.173*** 0.134*** 0.057** 0.001(0.095)
(0.047) (0.054) (0.027) (0.043)
Tariff reduction ×B2C sector 0.306**(0.142)
Control variables Yes Yes Yes Yes YesYear fixed effects Yes Yes
Yes Yes YesR-squared 0.13 0.05 0.08 0.06 0.08Observations 508 508
508 508 508
Standard errors are in parentheses. All tests are
two-tailed.*p< 0.10; ** p< 0.05; *** p< 0.01.
B2C sector
The arguments provided so far indicate that, whenfaced with
fiercer competition from abroad, U.S.companies increase their CSR
to improve their com-petitiveness and differentiate themselves from
theirforeign rivals. That being said, the value of CSR asa
differentiation strategy likely varies across busi-ness sectors. In
particular, Lev et al. (2010) showthat individual consumers are
more responsive tocompanies’ CSR engagement than industrial
buy-ers, which reflects inherent differences in the pur-chasing
decision-making process (Corey, 1991).20
Since sensitivity to CSR is likely higher for indi-vidual
customers, it follows that the differentiationgains from CSR should
be higher for companiesselling to individual customers (i.e., B2C
compa-nies), as opposed to companies selling to industrialbuyers.
Consequently, I should observe a strongertreatment effect for
companies in the B2C sector.
I examine this mechanism in Model 1, whereI augment my baseline
specification by includingan interaction term between the tariff
reductiondummy and a dummy variable indicating whethera company
operates in the B2C sector. The clas-sification of B2C industries
is obtained from Levet al. (2010: 188). As is shown, the
treatmenteffect is significantly stronger for companies in
20 More precisely, “[t]he purchasing decision of an individ-ual
consumer is affected not only by product attributes, butalso by
social group forces, psychological factors, and the con-sumer’s
situational forces. In contrast, in industrial purchasing,the
decision-making process is highly formalized, using
definedprocurement procedures, and subject to economic
(cost/value)analysis.” (Lev et al., 2010: 186, adapted from Corey,
1991)
the B2C sector, consistent with the
differentiationmechanism.
CSR dimensions
CSR initiatives can take on many different forms.For example,
companies may decide to invest in theresearch and development of
environment-friendlyproducts, offer work-life benefits (e.g., child
care,flextime) to their employees, donate to charity,etc. Given the
wide variety of CSR investments,their contribution to a company’s
competitivenessmay differ. More specifically, a company’s
socialengagement that directly addresses the needs ofits core
stakeholders (e.g., employees and con-sumers) may allow companies
to improve theircompetitiveness more effectively than social
activ-ities that are primarily directed at other, moreperipheral
stakeholders (e.g., society at large andenvironment).
For instance, CSR programs targeted at improv-ing product
quality may benefit domestic compa-nies in two ways. On one hand,
they may reducethe price elasticity of demand—consumers are
will-ing to pay a higher price for “ethical” goods.On the other
hand, they may increase consumerdemand directly by enhancing
consumer loyaltyand advocacy as well as attracting new
customerssuch as “green” consumers or, more generally, con-sumers
who are responsive to sustainable practices(see, e.g., Baron, 2008;
Du, Bhattacharya, and Sen,2007; Kotler, Hessekiel, and Lee, 2012;
Luo andBhattacharya, 2006; McWilliams and Siegel, 2001;Reinhardt,
1998; Sen and Bhattacharya, 2001).Relatedly, having a strong
employee-related CSR
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Does Competition Foster CSR? 1481
program may help companies attract, motivate, andmaintain the
most talented employees in the indus-try, thus directly adding to
the firm’s competitive-ness (e.g., Albinger and Freeman, 2000;
Greeningand Turban, 2000; Turban and Greening, 1996).21
In Models 2–5, I extend my baseline specifica-tion to examine
different types of CSR investments.Specifically, I decompose the
KLD index into foursub-indices by adding up KLD strengths
pertainingto employees, customers, environment, and societyat large
(i.e., all remaining KLD strengths), respec-tively. As can be seen
from Model 2, companiessubstantially increase their
employee-relatedCSR following the treatment, which is inline with
the labor channel suggested above.Moreover, Model 3 shows that
companies increasetheir customer-related CSR, which lends
additionalsupport to the differentiation channel. Finally,
theestimates in Models 4 and 5 indicate that companiesare less
likely to increase their CSR efforts targetedat other, more
peripheral stakeholders.
DISCUSSION AND CONCLUSION
This paper examines whether foreign competi-tion affects CSR
investments of domestic com-panies. Extending existing theories, I
argue thatdomestic companies respond to fiercer competitionfrom
abroad by increasing their CSR, as they arekeen to leverage their
comparative advantage (intheir relationships with local
stakeholders such asconsumers, employees, and communities) to
dif-ferentiate themselves and remain competitive. Toempirically
test this theoretical prediction, I exploita quasi-natural
experiment in the form of largeimport tariff reductions that
occurred between 1992and 2005 in the U.S. manufacturing sector.
Using amatched difference-in-differences approach, I findthat,
following the tariff reductions, domestic com-panies increase their
CSR efforts, as measured bya significant increase in their KLD
index. Thisresult is consistent with the view that CSR
generatesvaluable resources that allow domestic companies
21 Anecdotal evidence further supports these arguments: inthe
aforementioned survey by Accenture and UNGC (2010:14), “58% of CEOs
identify consumers as the most importantstakeholder group that will
impact the way they manage societalexpectations. Employees were
second with 45%.” Along similarlines, Jim Sinegal, Costco’s CEO,
argues: “I happen to believe thatin order to reward the shareholder
in the long term, you have toplease your customers and workers”
(Wall Street Journal, 2004).
to improve their competitiveness and differentiatethemselves
from their foreign rivals.
This finding is related to the economics literaturethat examines
the impact of globalization on socialand environmental welfare. In
particular, Copelandand Taylor (1994) argue that multinational
firmsmay exploit “pollution havens” in foreign countriesby, e.g.,
moving parts of their (pollution-intensive)production abroad to
countries with lax environ-mental standards. Yet, the empirical
literature findslittle empirical evidence that trade has a
detrimen-tal effect on the environment globally (Eskelandand
Harrison, 2003; Frankel and Rose, 2005; Gross-man and Krueger,
1993, 1995). Similarly, whileit is sometimes argued that
globalization increasesthe incidence of child labor, the empirical
evidenceseems to suggest that trade openness may in factreduce
child labor (Edmonds and Pavcnik, 2005;Neumayer and De Soysa,
2005). A common featureof these articles is the focus on aggregate
social andenvironmental welfare. In contrast, my paper stud-ies
firm-level responses to trade liberalization froma strategic CSR
perspective.
Furthermore, this paper contributes to theliterature on product
market competition andCSR. The papers that are most closely
relatedare Fernandez-Kranz and Santalo (2010), Fismanet al. (2006),
and Declerck and M’Zali (2012).Consistent with my findings, they
find a positivecorrelation between competition (proxied by theHHI
of industry concentration) and CSR. However,as mentioned in the
introduction, such correlationdoes not warrant a causal
interpretation. Severalunobserved variables may correlate with both
HHIand CSR, and hence drive a spurious relationshipbetween the two.
To the best of my knowledge, mypaper is the first to examine the
causal effect ofproduct market competition on CSR.
A potential limitation of my study is that,although it shows
that U.S. companies respond to areduction in import tariffs by
increasing their socialengagement, it does not provide direct
evidencethat this increase in CSR is value enhancing.An alternative
interpretation of my results couldbe that fiercer competition leads
to corporateinefficiencies that translate into wasteful CSRefforts.
Nevertheless, this alternative interpretationis very unlikely, for
two reasons. First, if—as manyeconomists argue—product market
competitionfosters efficiency (e.g., Alchian, 1950; Friedman,1953;
Stigler, 1958), it seems implausible thatcompanies would respond to
higher competition by
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
1482 C. Flammer
increasing CSR if doing so were value destroying.Second, a large
literature examines the relationshipbetween CSR and financial
performance (forreviews, see, e.g., Margolis, Elfenbein, and
Walsh,2007; Margolis and Walsh, 2001, 2003; Orlitzky,Schmidt, and
Rynes, 2003). While there is someheterogeneity in the findings,
Margolis et al. (2007)note in their meta-analysis of this
literature that the“overall effect is positive but small” (p. 2).
Thissuggests that CSR is unlikely to destroy value.22
A caveat of my study is that it is empiricallydifficult to
provide evidence on the causal mecha-nisms through which CSR may
improve U.S. com-panies’ ability to compete with their foreign
rivals.In auxiliary analyses, I provide evidence that isindicative
of potential mechanisms, yet this evi-dence is merely suggestive as
alternative interpre-tations cannot be ruled out. For instance, I
showthat the treatment effect is stronger in the B2Csector. To the
extent that individual customers aremore sensitive to companies’
CSR engagementthan industrial buyers (Lev et al., 2010), this
evi-dence is potentially consistent with the differentia-tion
mechanism. However, it is open to alternativeinterpretations—e.g.,
companies making goods inthe B2C sector may respond more strongly
sim-ply because foreign competitors are more likely tobe the
low-cost producers for these goods. Moregenerally, this illustrates
the caveat of using inter-action terms in a
difference-in-differences setting.While the treatment effect (i.e.,
the effect of importtariff reductions on CSR) is well identified,
thismay not be the case of the interaction effects,since they are
obtained by interacting the treat-ment dummy with cross-sectional
characteristicsfor which I do not have exogenous variation
(e.g.,being in the B2C sector is not exogenously deter-mined, and
hence may correlate with unobservablecharacteristics that may also
explain the hetero-geneity in the treatment effect). Relatedly, my
find-ing that companies increase their employee-relatedKLD
strengths is suggestive of a labor productiv-ity mechanism. Yet, as
KLD strengths pertainingto employees include a broad list of
criteria (e.g.,work/life benefits, gay and lesbian policies,
unionrelations, health and safety, employee involvement,
22 A caveat of this literature is that CSR is endogenous
withrespect to financial performance. However, recent evidence
byFlammer (2014), who relies on exogenous variation in CSR inthe
form of CSR-related shareholder proposals that pass or fail bya
small margin of votes, suggests that the positive link betweenCSR
and financial performance is in fact causal.
stock ownership, etc.), alternative interpretationscannot be
ruled out. For instance, employee involve-ment in decision making
and stock ownership couldbe interpreted as devices to adapt or gain
employeesupport for change (see, e.g., Morgan and Zeffane,2010;
Piderit, 2000). As these examples illustrate,providing conclusive
evidence on the underlyingmechanisms is a challenging task that
would requiredetailed microdata on the companies’ operationsand
processes. Making ground on these mecha-nisms is an exciting avenue
for future research.
My findings have several managerial impli-cations. First, the
fact that domestic companiesrespond to import tariff cuts by
increasing theirCSR suggests that CSR helps companies
remaincompetitive and differentiate themselves from theirforeign
rivals. Hence, in the face of rising globalcompetition, managers
may find it worthwhile todesign and implement effective CSR
practices.Second, my findings suggest that CSR is partof a firm’s
competitive strategy, and hence maybe more core to corporate
strategy than oftenthought. Accordingly, managers could benefit
fromexplicitly integrating social and environmentalconsiderations
into their strategic decision making.
Finally, finding that the lowering of trade barriersfosters
domestic companies’ CSR has potentiallyimportant policy and welfare
implications. In theeconomics literature, the typical view is that
tradeliberalization increases social surplus by improv-ing
productive efficiency and consumers’ welfare.The results of this
study suggest that the welfare ofthe companies’ stakeholders
(including consumers,employees, and the environment) improves as
well.Accordingly, taking into account this positive exter-nality,
the overall benefits of trade liberalization onsociety may be
larger than previously assumed.
ACKNOWLEDGEMENTS
I thank Constance Helfat (the editor), two anony-mous reviewers,
Gautam Ahuja, Tima Bansal,Simon Johnson, Aleksandra Kacperczyk, S.
P.Kothari, Anita McGahan, Lamar Pierce, BrianRichter, Arvind
Subramanian, Paul Vaaler, con-ference participants at the
Sustainability and theCorporation: Big Ideas Conference
(HarvardBusiness School), the SMS Special Conference:Startup and
Restart Strategies (Tel Aviv), the 73rdAnnual Meeting of the
Academy of Management(Orlando, FL), the 13th Annual Strategy and
the
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Does Competition Foster CSR? 1483
Business Environment Conference (UT Austin), aswell as seminar
participants at the University ofMinnesota, Ivey, INSEAD, HEC
Paris, HEC Lau-sanne, Baruch, and Bentley for valuable commentsand
suggestions.
REFERENCES
Accenture and UNGC. 2010. A New Era of Sustainabil-ity: UN
Global Compact-Accenture CEO Study 2010.United Nations Global
Compact and Accenture: NewYork.
Agle BR, Donaldson T, Freeman RE, Jensen MC,Mitchell RK, Wood
DJ. 2008. Dialogue: towardssuperior stakeholder theory. Business
Ethics Quarterly18(2): 153–190.
Albinger HS, Freeman SJ. 2000. Corporate social perfor-mance and
attractiveness as an employer to different jobseeking populations.
Journal of Business Ethics 28(3):243–253.
Alchian AA. 1950. Uncertainty, evolution, and economictheory.
Journal of Political Economy 58(3): 211–221.
Almeida H, Campello M, Laranjeira B, Weisbenner S.2012.
Corporate debt maturity and the real effects of the2007 credit
crisis. Critical Finance Review 1(1): 3–58.
Barney J. 1991. Firm resources and sustained
competitiveadvantage. Journal of Management 17(1): 99–120.
Baron DP. 2008. Managerial contracting and corporatesocial
responsibility. Journal of Public Economics92(1/2): 268–288.
Berman SL, Wicks AC, Kotha S, Jones TM. 1999.Does stakeholder
orientation matter? The relationshipbetween stakeholder management
models and firmfinancial performance. Academy of Management
Jour-nal 42(5): 488–506.
Bernard AB, Jensen JB, Schott PK. 2006a. Trade costs,firms, and
productivity. Journal of Monetary Eco-nomics 53(5): 917–937.
Bernard AB, Jensen JB, Schott PK. 2006b. Survival ofthe best
fit: exposure to low-wage countries and the(uneven) growth of U.S.
manufacturing plants. Journalof International Economics 68(1):
219–237.
Clausing KA. 2001. Trade creation and trade diversionin the
Canada-United States Free Trade Agreement.Canadian Journal of
Economics 34(3): 677–696.
Copeland BR, Taylor MS. 1994. North-South trade and
theenvironment. Quarterly Journal of Economics 109(3):755–787.
Corey ER. 1991. Industrial Marketing Cases and Con-cepts.
Prentice Hall: Englewood Cliffs, NJ.
Davis K. 1973. The case for and against business assump-tion of
social responsibilities. Academy of ManagementJournal 16(2):
312–323.
Deckop JR, Merriman KK, Gupta S. 2006. The effects ofCEO pay
structure on corporate social performance.Journal of Management
32(3): 329–342.
Declerck MD, M’Zali B. 2012. Product market competi-tion and
corporate social responsibility. Working paper,University Lille
Nord de France, Lille, France.
Domini Social Investments. 2013. Fund Fact Sheet. DSILInvestment
Services: New York.
Du S, Bhattacharya CB, Sen S. 2007. Reaping relationalrewards
from corporate social responsibility: the roleof competitive
positioning. International Journal ofResearch in Marketing 24(3):
224–241.
Edmonds EV, Pavcnik N. 2005. The effect of trade liber-alization
on child labor. Journal of International Eco-nomics 65(2):
401–419.
Elhauge E. 2005. Sacrificing corporate profits in the pub-lic
interest. New York University Law Review 80(3):733–869.
Eskeland GS, Harrison AE. 2003. Moving to greener pas-tures?
Multinationals and the pollution haven hypothe-sis. Journal of
Development Economics 70(1): 1–23.
Fan J, Lang L. 2000. The measurement of relatedness:an
application to corporate diversification. Journal ofBusiness 73(4):
629–660.
Feenstra RC. 1996. U.S. imports, 1972–1994: data
andconcordances. NBER Working paper 5515, NationalBureau of
Economic Research: Cambridge, MA.
Feenstra RC. 1998. Integration of trade and disintegrationof
production in the global economy. Journal of Eco-nomic Perspectives
12(4): 31–50.
Feenstra RC, Romalis J, Schott PK. 2002. U.S. imports,exports,
and tariff data, 1989–2001. NBER Workingpaper 9387, National Bureau
of Economic Research:Cambridge, MA.
Fernandez-Kranz D, Santalo J. 2010. When necessitybecomes a
virtue: the effect of product market com-petition on corporate
social responsibility. Journal ofEconomics and Management Strategy
19(2): 453–487.
Fisman R, Heal G, Nair VB. 2006. A model of
corporatephilanthropy. Working paper, Columbia University,New
York.
Flammer C. 2013. Corporate social responsibility andshareholder
reaction: the environmental awareness ofinvestors. Academy of
Management Journal 56(3):758–781.
Flammer C. 2014. Does corporate social responsibilitylead to
superior financial performance? A regressiondiscontinuity approach.
Management Science, forth-coming.
Forbes. 2011. Seventh generation’s new CEO onwhy sustainability
is necessary. Forbes CSR Blog.9 February. Available at:
(http://www.forbes.com/sites/csr/2011/02/09/seventh-generations-new-ceo-on-why-sustainability-is-necessary/).
Frankel JA, Romer D. 1999. Does trade cause growth?American
Economic Review 89(3): 379–399.
Frankel JA, Rose AK. 2005. Is trade good or bad forthe
environment? Sorting out the causality. Review ofEconomics and
Statistics 87(1): 85–91.
Freeman RE. 1984. Strategic Management: A StakeholderApproach.
Pitman: Boston, MA.
Fresard L. 2010. Financial strength and product marketbehavior:
the real effects of corporate cash holdings.Journal of Finance
65(3): 1097–1122.
Fresard L, Valta P. 2014. How does corporate investmentrespond
to increased entry threat? HEC Paris ResearchPaper FIN-2014-1046,
HEC Paris, Paris, France.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
1484 C. Flammer
Friedman M. 1953. The methodology of positive eco-nomics. In
Essays in Positive Economics. University ofChicago Press: Chicago,
IL, 3–43.
Friedman M. 1962. Capitalism and Freedom. University ofChicago
Press: Chicago, IL.
Friedman M. 1970. The social responsibility of business isto
increase its profits. The New York Times Magazine 13September:
SM17.
Frye T, Mansfield ED. 2004. Timing is everything: elec-tions and
trade liberalization in the postcommunistworld. Comparative
Political Studies 37(4): 371–398.
Graves SB, Waddock SA. 1994. Institutional owners andcorporate
social performance. Academy of ManagementJournal 37(4):
1034–1046.
Greening DW, Turban DB. 2000. Corporate social perfor-mance as a
competitive advantage in attracting a qualityworkplace. Business
& Society 39(3): 254–280.
Gros D. 1987. A note on the optimal tariff, retaliation,and the
welfare loss from tariff wars in a frameworkwith intra-industry
trade. Journal of International Eco-nomics 23(3/4): 357–367.
Grossman GM, Helpman E. 1995. The politics offree-trade
agreements. American Economic Review85(4): 667–690.
Grossman GM, Krueger AB. 1993. Environmental impactsof a North
American Free Trade Agreement. In TheMexico-U.S. Free Trade
Agreement. MIT Press: Cam-bridge, MA; 13–55.
Grossman GM, Krueger AB. 1995. Economic growthand the
environment. Quarterly Journal of Economics110(2): 353–377.
Hart SL. 1995. A natural resource-based view of the firm.Academy
of Management Review 20(4): 986–1014.
Heckscher EF. 1919. The effect of foreign trade on the
dis-tribution of income [in Swedish]. Ekonomisk Tidskrift21(2):
1–32. Reprinted in Readings in the Theory ofInternational Trade.
Irwin: Homewood, IL, 272–300.
Helfat CE, Peteraf MA. 2003. The dynamic resource-basedview:
capability lifecycles. Strategic ManagementJournal 24(10):
997–1010.
Helpman E, Krugman PR. 1989. Trade Policy and MarketStructure.
MIT Press: Cambridge, MA.
Henisz WJ, Mansfield ED. 2006. Votes and vetoes: thepolitical
determinants of commercial openness. Inter-national Studies
Quarterly 50(1): 189–212.
Hiscox MJ. 2002. International Trade and Political Con-flict:
Commerce, Coalitions, and Mobility. PrincetonUniversity Press:
Princeton, NJ.
Hollander M, Wolfe DA. 1999. Nonparametric StatisticalMethods
(2nd edn). Wiley Series in Probability andStatistics: New York.
Hooley G, Broderick A, Moeller K. 2006. Competitivepositioning
and the resource-based view of the firm.Journal of Strategic
Marketing 6(2): 97–113.
Johnson-Cramer ME. 2004. Organization-level antece-dents of
stakeholder conflict: a comparative case study.Academy of
Management Proceedings: F1–F6.
Jones TM. 1995. Instrumental stakeholder theory: a syn-thesis of
ethics and economics. Academy of Manage-ment Review 20(2):
404–437.
Kacperczyk A. 2009. With greater power comes
greaterresponsibility? Takeover protection and corporate
attention to stakeholders. Strategic ManagementJournal 30(3):
261–285.
Kotler P, Hessekiel D, Lee N. 2012. Good Works!: Market-ing and
Corporate Initiatives that Build a Better World… and the Bottom
Line. Wiley: Hoboken, NJ.
Krugman PR. 1995. Growing world trade: causes andconsequences.
Brookings Papers on Economic Activity26(1): 327–377.
Krugman PR, Obstfeld M, Melitz M. 2012. InternationalEconomics:
Theory and Policy (9th edn). Pearson Press:Upper Saddle River,
NJ.
Lev B, Petrovits C, Radhakrishnan S. 2010. Is doing goodgood for
you? How corporate charitable contributionsenhance revenue growth.
Strategic Management Jour-nal 31(2): 182–200.
Lileeva A, Trefler D. 2010. Improved access to for-eign markets
raises plant-level productivity … forsome plants. Quarterly Journal
of Economics 125(3):1051–1099.
Luo X, Bhattacharya CB. 2006. Corporate social responsi-bility,
customer satisfaction, and market value. Journalof Marketing 70(4):
1–18.
Margolis JD, Elfenbein HA, Walsh JP. 2007. Does it payto be
good? A meta-analysis and redirection of researchon the
relationship between corporate social and finan-cial performance.
Working paper, Harvard BusinessSchool: Boston, MA.
Margolis JD, Walsh JP. 2001. People and Profits: TheSearch for a
Link Between a Company’s Social andFinancial Performance. Lawrence
Erlbaum: Mahweh,NJ.
Margolis JD, Walsh JP. 2003. Misery loves companies:rethinking
social initiatives by business. AdministrativeScience Quarterly
48(2): 268–305.
Mattingly JE, Berman SL. 2006. Measurement of corpo-rate social
action: discovering taxonomy in the KinderLydenburg Domini ratings
data. Business & Society45(1): 20–46.
Mayer W. 1981. Theoretical considerations on negotiatedtariff
adjustments. Oxford Economic Papers 33(1):135–153.
McWilliams A, Siegel D. 2001. Corporate social respon-sibility:
a theory of the firm perspective. Academy ofManagement Review
26(1): 117–127.
MIT Sloan Management Review. 2012. Sustainability nearsa tipping
point. MIT Sloan Management Review 53(2):69–74.
Morgan D, Zeffane R. 2010. Employee involvement, orga-nizational
change and trust in management. Interna-tional Journal of Human
Resource Management 14(1):55–75.
Neumayer E, De Soysa I. 2005. Trade openness, foreigndirect
investment and child labor. World Development33(1): 43–63.
New York Times. 2011. First, make money. Also, do good.New York
Times 14 August: BU3.
Ohlin BG. 1933. Interregional and International Trade.Harvard
University Press: Cambridge, MA.
Orlitzky M, Schmidt FL, Rynes SL. 2003. Corporate socialand
financial performance: a meta-analysis. Organiza-tion Studies
24(3): 403–441.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Does Competition Foster CSR? 1485
Ornelas E. 2005. Endogenous free trade agreements andthe
multilateral trading system. Journal of InternationalEconomics
67(2): 471–497.
Peteraf MA. 1993. The cornerstones of competitive advan-tage: a
resource-based view. Strategic ManagementJournal 14(3):
179–191.
Piderit SK. 2000. Rethinking resistance and rec-ognizing
ambivalence: a multidimensionalview of attitudes towards an
organizationalchange. Academy of Management Review
25(4):783–794.
Pierce JR, Schott PK. 2012. The surprisingly swift declineof
U.S. manufacturing employment. NBER Workingpaper 18655, National
Bureau of Economic Research:Cambridge, MA.
Porter ME. 1991. America’s green strategy. ScientificAmerican
264(4): 168.
Porter ME, Kramer MR. 2006. Strategy & society: thelink
between competitive advantage and corporatesocial responsibility.
Harvard Business Review 84(12):78–92.
Porter ME, Kramer MR. 2011. The big idea: creatingshared value.
Harvard Business Review 89(1–2):62–77.
Reinhardt FL. 1998. Environmental product differenti-ation:
implications for corporate strategy. CaliforniaManagement Review
40(4): 43–73.
Ricardo D. 1817. The Principles of Political Economy
andTaxation. John Murray: London.
Rogowski R. 1989. Commerce and Coalitions: How TradeAffects
Domestic Political Alignments. Princeton Uni-versity Press:
Princeton, NJ.
Russo MV, Fouts PA. 1997. A resource-based perspec-tive on
corporate environmental performance andprofitability. Academy of
Management Journal 40(3):534–559.
Schott PK. 2010. U.S. manufacturing exports and importsby SIC or
NAICS category and partner country,1972–2005. Working paper, Yale
University: NewHaven, CT.
Sen S, Bhattacharya CB. 2001. Does doing good alwayslead to
doing better? Consumer reactions to corporatesocial responsibility.
Journal of Marketing Research38(2): 225–243.
Stigler GJ. 1958. The economics of scale. Journal of Lawand
Economics 1(1): 54–71.
Trefler D. 2004. The long and short of the Canada-U.S.Free Trade
Agreement. American Economic Review94(4): 870–985.
Turban DB, Greening DW. 1996. Corporate social perfor-mance and
organizational attractiveness to prospectiveemployees. Academy of
Management Journal 40(3):658–672.
Wall Street Journal. 2004. Costco’s dilemma: be kind to
itsworkers, or Wall Street? Wall Street Journal 26 March:B1.
Wall Street Journal. 2013. Not all sneakers are imported…yet.
Wall Street Journal 28 February: B1.
Wernerfelt B. 1984. A resource-based view of the firm.Strategic
Management Journal 5(2): 171–180.
Wernerfelt B. 1995. A resource-based view of the firm:ten years
after. Strategic Management Journal 16(3):171–174.
Wood A. 1995. North-South Trade, Employment andInequality:
Changing Fortunes in a Skill-Driven World.Oxford University Press:
Oxford, UK.
SUPPORTING INFORMATION
Additional supporting information may be foundin the online
version of this article:
Figure S1. Import tariff rates in control and treat-ment
group.Table S1. Industries affected by large import
tariffreductions.Table S2. Robustness.
Copyright © 2014 John Wiley & Sons, Ltd. Strat. Mgmt. J.,
36: 1469–1485 (2015)DOI: 10.1002/smj
-
Figure S1. Import tariff rates in control and treatment
group
-0.0150
-0.0100
-0.0050
0.0000
0.0050
0.0100
0.0150
0.0200
0.0250
0.0300
0.0350
-5 -4 -3 -2 -1 1 2 3 4 5
Tarif
f rat
e
Year relative to treatment
Treatment group
Control group
Difference between treatment and control group (95% confidence
interval within dotted lines)
-
Table S1. Industries affected by large import tariff
reductions
Year SIC Industry description Multilateral agreement
1992 3613 Switchgear and switchboard apparatus Other1992 3669
Communications equipment, nec Other1993 2761 Manifold business
forms GATT1993 2522 Office furniture, except wood GATT, NAFTA1993
2451 Mobile homes GATT, NAFTA1993 3715 Truck trailers GATT,
NAFTA1994 3651 Household audio and video equipment Other1994 3577
Computer peripheral equipment, nec GATT, NAFTA1994 3341 Secondary
nonferrous metals GATT1995 3555 Printing trades machinery WTO,
NAFTA1995 2834 Pharmaceutical preparations WTO 1995 2835 Diagnostic
substances WTO1995 3822 Environmental controls Other1995 3944
WTO1995 3011 Tires and inner tubes WTO1995 3842 Surgical appliances
and supplies WTO1995 2842 Polishes and sanitation goods WTO,
NAFTA1995 3579 Office machines, nec Other1995 2844 Toilet
preparations Other1995 3942 Dolls and stuffed toys WTO1995 2833
Medicinals and botanicals WTO1995 3559 Special industry machinery,
nec WTO, NAFTA1995 3612 Power, distribution and specialty
transformers Other1995 3843 Dental equipment and supplies WTO1995
3561 Pumps and pumping equipment Other1997 3695 Magnetic and
optical recording media WTO, NAFTA1997 3812 Search and navigation
equipment WTO, NAFTA1997 3578 Calculating and accounting equipment
WTO, NAFTA1997 3826 Analytical instruments WTO, NAFTA1997 3844
X-ray apparatus and tubes WTO, NAFTA1998 3829 Measuring and
controlling devices, nec Other1998 3845 Electromedical equipment
Other1998 3089 Plastics products, nec Other1998 3663 Radio and T.V.
communications equipment Other
-
Table S2. Robustness
Tariff reductions Tariff reductions Matching based Matching
based Matching based Excluding related Tariff reductions
Controlling for > 2 × cutoff > 4 × cutoff on characteristics
on 3-digit SIC on location industries from due to GATT, changes
in
at t – 3 industries matched sample WTO, or NAFTA advertising
Dependent variable: KLD KLD KLD KLD KLD KLD KLD KLD
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model
8
Tariff reduction 0.251*** 0.504*** 0.393*** 0.394*** 0.436***
0.507*** 0.301** 0.360***(0.059) (0.151) (0.095) (0.092) (0.115)
(0.092) (0.122) (0.085)
Control variables Yes Yes Yes Yes Yes Yes Yes YesYear fixed
effects Yes Yes Yes Yes Yes Yes Yes Yes
R-squared 0.12 0.14 0.12 0.17 0.14 0.14 0.16 0.13Observations
1,092 222 460 414 286 508 284 508
All tests two-tailed. * p < 0.10; ** p < 0.05; *** p <
0.01.
Flammer-2015-Strategic_Management_JournalAppendix